TaxoLogic - Explainable Hybrid Artificial Intelligence for Transfer Pricing

TaxoLogic develops a hybrid experimental laboratory artificial intelligence (AI) framework that combines data-driven learning with logic-based reasoning to make regulatory decision processes transparent, explainable, and auditable. In tax law, current AI tools act largely as opaque systems—efficient in processing vast information but unable to justify their conclusions. This lack of transparency poses major risks in domains such as tax law, where every decision must be consistent, traceable, and legally defensible. TaxoLogic bridges this gap by integrating the linguistic power of large language models (LLMs) with symbolic reasoning methods that ensure logical consistency, auditability, and adherence to legal standards.

TaxoLogic focuses on transfer pricing, a central challenge for multinational enterprises in determining cross-border transaction values. The task requires the interpretation of overlapping and often contradictory sources—OECD guidelines, national laws, and bilateral treaties. TaxoLogic transforms these texts into structured, machine-readable rules and applies logic-based reasoning to derive coherent and verifiable interpretations. The framework can automatically generate both compliant results and transparent explanations of how these outcomes were reached, supporting auditors, regulators, and tax experts alike.

The main objectives are to link natural-language interpretation with formal reasoning, establish a hybrid AI model for reliable legal compliance, validate the approach in real-world transfer pricing cases, and produce reusable research assets such as rule-extraction pipelines, logic frameworks, and benchmark datasets. TaxoLogic also promotes European technological sovereignty by prioritizing open and verifiable AI components and aims to strengthen trust in AI-based governance through transparency and explainability.

TaxoLogic unites eight complementary partners from academia and industry: FH Steyr and WU Vienna contribute legal and tax expertise, while the Research Institute for Symbolic Computation at JKU and RISC Software GmbH provide core research on symbolic reasoning and hybrid architectures. Rise2Reality serves as a key technical and architectural partner, focusing on benchmarking LLMs for regulatory language processing and the practical productization of the hybrid AI framework, and the industrial partners—MIC, voestalpine, and Greiner—supply real data, validation scenarios, and practical insight from global trade, supply chain, and corporate tax compliance. This combination ensures that TaxoLogic’s outcomes are both scientifically sound and directly applicable in operational contexts.

After the project, results will evolve into pilot applications and commercial products integrating hybrid AI into tax and compliance systems, increasing efficiency and reducing risks. The project promotes sustainable digital transformation by automating manual processes, improving decision quality, and ensuring explainable, inclusive AI accessible to diverse professionals.